As it turns out, the cib.xml file had some failures that were not showing up
in crm_mon.  I manually removed them by issuing cibadmin -D -X '<id>' and
then tried again, and now it is working the way it should when it comes to
switching over the resources.  I knew it was right under my nose!

I was lead into the right direction though, and for that, I am thankful for
you who are more knowledgeable.


Christian Iversen wrote:
> 
> Dejan Muhamedagic wrote:
>> Hi,
>> 
>> On Thu, Apr 16, 2009 at 01:15:55PM -0700, Ethan Bannister wrote:
>>> /var/log/messages on san2 states that it couldn't promote drbd1:1 on
>>> san2
>>> because san1 was still in primary mode.  This makes sense.  But why
>>> would it
>>> have no issues with taking down the other drbd devices on san1 and not
>>> drbd1?  Is there a log file that may give me a better idea of what may
>>> be
>>> going on?  I am assuming that when I pull the cable or take down eth0,
>>> the
>>> rest of the cluster is unable to tell san1 to demote the drbd devices so
>>> that san2 can then promote them.  But from what I gather from this log
>>> file,
>>> drbdadm does all of this.  So would it be safe to assume that drbdadm
>>> communicates through the direct link between the two targets and it is
>>> failing for drbd1 for some reason?
>> 
>> AFAIK, drbd is using just one link. If that cable is pulled, then
>> you have a drbd split brain. BTW, you may want to take a look at
>> dopd to have heartbeat help drbd in this case.
> 
> In a word, no. DRBD uses an explicit master/slave system, so the nodes 
> will always pre-agree on who will become the master. In a disconnect 
> situation, the nodes will go into master/UNKNOWN and UNKNOWN/slave, 
> respectively.
> 
> We implemented a 2-node storage cluster with drbd and heartbeat. We use 
> ipfail to support 2 links for IP-failover, but drbd only requires the 
> one connection.
> 
> -- 
> Med venlig hilsen
> Christian Iversen
> _______________________________________________
> Linux-HA mailing list
> [email protected]
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
> 
> 

-- 
View this message in context: 
http://www.nabble.com/DRBD-does-not-switch-resources-to-other-node-properly-tp23082432p23105594.html
Sent from the Linux-HA mailing list archive at Nabble.com.

_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to